Coordination of verbal dependents in Old French:
Coordination as a specified juxtaposition or apposition
Nicolas Mazziotta
Universität Stuttgart, Institut für Linguistik/Romanistik
[email protected]
Abstract
amount of money to Sir W.” (Charter,
1278, 8)
Scholars have proposed many different
models to describe coordination of verbal
dependents. We give a brief presentation
of the most common ways to deal with
this construction from a general point of
view. Then, we evaluate the adequacy of
the models using data from Old French. In
this particular language, coordination is a
more elaborated form of juxtaposition and
apposition, which differs only at the semantic level. For this reason, the coordinating conjunction has to be considered as
a dependent of the following conjunct.
This phenomenon is named pairs of synonyms (Fr.
binôme synonymiques), and the link between this
kind of structure and translations in the Middle
Ages has often been studied from the perspective
of stylistics. The semantic relation between the
synonyms varies, and it is generally assumed that
pairs of synonyms are used for the sake of clarity
(Buridant, 1977; Buridant, 1980). Buridant (1977,
294, our translation) proposes the following definition:
a sequence of two synonyms normally
belonging to the same part of speech and
sharing the same level in the syntactic
hierarchy
Introduction
Our purpose is to present an adequate way to describe simple coordination of verbal dependents in
Old French (hereafter “OF”) within a dependency
framework. We will mainly focus on the question
of the hierarchical position of the conjunction.
As far as coordination constructions are concerned, OF is not very different from modern
European languages, such as English or modern
French. However, some uses of the conjunction
et in OF would not be possible nowadays. For
example, the construction cel pris et celle summe
d’argent in ex. 1 would be ungrammatical in modern French (or English), because both nouns refer
to the same object, and modern French does not
allow the coordination of two noun phrases with
identical referents.
We would like to compare this kind of coordination with cases that can be analysed in the same
way as modern variants, and to propose an adequate and accurate hierarchy to model them. The
focus of our presentation will gradually shift from
general considerations about coordination toward
specific OF properties.
We begin this paper (section 1) with a review of
the main descriptive options that have been used
to analyse coordination in a dependency framework. In section 2, we briefly highlight the fact
that OF sentences can often be grammatically correct without the use of segmental grammatical devices such as prepositions and conjunctions. In
section 3, we survey OF juxtaposition and apposition. We provide evidence that both constructions can be syntactically and semantically complemented by the use of the same conjunction – a
process very close to the one called specification
by Lemaréchal (1997) – thus forming two different kinds of coordination.
(1) cel pris et celle summe d’argent
this price and this amount of money
doit li glise Saint-Donis paier a
must the church S-D
pay to
mun saingor Wilhame
my sir W.
“Saint Denis church owes this price and
28
1
fall
Coordination in the dependency
framework
Alfred — and — Bernard
To begin with, we provide a general overview of
models of coordination in the dependency framework. Since the concept of dependency varies
among theories, we will briefly introduce the different definitions when necessary. We illustrate
this section with English translations for the sake
of simplicity. We conclude section 1 with a summary of the descriptive options provided by these
different models. The appropriate formalism to
model OF coordination will be elaborated in the
following sections.
1.1
Figure 1: Coordination according to Tesnière
When the conjunction is not present, the representation is exactly the same, except the horizontal
line is unbroken. Tesnière’s model of coordination
multiplies the number of dependents that can be
connected to a verb.
1.2
Mel’čuk’s unidimensional approach
In the Meaning-Text Theory (MTT) framework,
coordination is described as a dependency relation.
MTT has developed a comprehensive list of criteria to find syntactic dependencies, to identify
the governor in such a relation, and to classify
them (Mel’čuk, 2009, 25-40). To identify a governor, syntactic (with higher priority), morphological and semantic (with lower priority) aspects have
to be investigated. Syntactically, the passive valency of the phrase formed by the governor and its
dependents should lead us to identify the governor
of the phrase. Morphologically, the governor controls agreement between the phrase and its context. Semantically, the governor is a better sample
of the referential class denoted by the phrase (e.g.:
a ham sandwich is a kind of sandwich, therefore,
ham is the dependent).
In fact, Mel’čuk (2009, 50-51) defines coordination from both a semantic and a syntactic perspective: no conjunct semantically depends on the
other, but the second conjunct syntactically depends on the first one. Coordination often uses a
conjunction and displays the following properties
(Mel’čuk, 1988, 41):
Tesnière’s baseline
Lucien Tesnière (1965, ch. 134 sqq.) introduces
the concept of jonction (we use the translation
junction hereafter), used to model coordination.
Junction is a “horizontal” relation. Words linked
in junction are hierarchically equivalent (Tesnière,
1965, ch. 135). This characteristic makes junction
very different from connexion (fr. connection),
which represents a governor/dependent “vertical”
relation, where the governor (the top node in the
stemma) is hierarchically more prominent than the
dependent. Dependency as such is never defined
by Tesnière, but Garde (1981, 159-160), in the
same framework, defines the governor as the word
that controls the passive valency of the phrase (the
potential it has to be dependent on some external
governor).
As a simple example of junction, we can analyse ex. 2: see fig. 1 (Tesnière, 1965, ch. 136, §3).
(2) Alfred and Bernard fall (translation of
stemma 248 in Tesnière’s book)
As the graphical (bi-dimensional) representation is
very important to him, Tesnière adds (we will see
in section 3.2 how this compares with the way appositions are handled):
1. In a phrase of the form X and Y, no
element can remain “independent”,
i.e., unrelated to any other element.
[...]
2. In the phrase X and Y, the conjunction cannot be the head, since the
distribution of the phrase is determined by its conjuncts and by no
means by the conjunction. [...]
3. X is the head of the phrase, since
the distribution of X and Y is that
of X, and by no means that of and
Y.
Two joined nodes each retain equivalent
vertical connections [i.e. dependency].
As a result, the graphical representation
derived from two vertical connections
and the junction line will always form
a triangle. (Tesnière, 1965, ch. 136, §4,
our translation)
Graphically, the conjunction and is placed directly
on the horizontal line.
29
grouping of sub-bubbles without any dependency
relation between them. The advantage of this
model is that it can cope with gapping and valency
slot coordination, but our main interest is the hierarchical position of the conjunction. In the representation shown in fig. 3,
4. In the chunk and Y, the conjunction
is the head: it determines the distribution of the expression to a greater
degree than Y. [...]
As a result, the analysis (see fig. 2) forces the
dependency between Bernard and fall to become
indirect, which was not the case with Tesnière’s
model.
fall
Alfred
fall
and
Bernard
Alfred
Figure 3: Coordination in a Bubble-tree
and
it can be seen that the representation leaves the exact hierarchical position of the coordinating conjunction unspecified: it is simply a sibling of the
conjuncts. Note that the dependency links the
whole bubble to its governor, thus assuming functional equivalence of the conjuncts.
Bernard
Figure 2: Coordination according to the MTT
According to the MTT, coordination can be direct, and it corresponds to traditional juxtaposition.
The author himself aknowledges that his puredependency model cannot describe constituent coordination efficiently (Mel’čuk, 2009, 93). For
instance, there is no difference in the description
of old men and women meaning “old men + old
women” and “old men + women (either old or
not)” (Mel’čuk, 2009, 93). Another limit of the
formalism appears in gapping coordinations or valency slot coordinations (non-constituent coordination). There is no way to correctly describe clustering as observed in: John loves Mary; and Peter,
Ann and John gets a letter from Mary and roses
from Ann.
1.3
Paradigmatic piles. The so-called paradigmatic pile device is aimed at easing transcription
and analysis of oral performance, mainly to deal
with disfluencies and reformulations. It inherits
the ideas of the grid analysis (Blanche-Benveniste
and Jeanjean, 1987, 167-171). Kahane and Gerdes
(2009) argue that the same device can be used to
describe coordination and apposition – the same
idea already appears in Bilger (1999), but without further formalisation. For instance, the following example presents a disfluency (Kahane and
Gerdes, 2009, § 3.2):
(3) okay so what what changed your mind
what and what . . . mind form some kind of
paradigm. Production is indeed interrupted, and
one could not reasonably think that both elements
are part of the same syntactic structure; as far as
reformulation and coordination are concerned,
Two dimensional formalisms
It is a common idea that the limits of the MTT
syntactic description of coordination are linked to
the unidimensionality of the formalism (generally
called projectivity). However, as Kahane (1997,
§ 5.5) states,
we consider that a segment Y of an utterance piles up with a previous segment
X if Y fills the same syntactic position as
X. (Kahane and Gerdes, 2009, § 4)
Subordination and coordination are two
orthogonal linguistic operations and we
need a two dimensional formalism to
capture this [. . . ]
Such an analysis is represented in fig. 4, where
curly brackets delimit the pile, and the vertical bar
divides the elements of the pile.
Besides, paradigmatic piles can also be used to
sketch a coordination relation: the analysis of ex. 2
is shown in fig. 5, where the italicised and is called
a pile marker. It is related to the conjuncts, but
their exact dependency is not stated:
Bubbles. Kahane (1997) introduces the concept
of the bubble. Bubbles are formal objects that represent embeddable clusters of nodes. Clustered elements are linked together by a dependency (this
concept is defined formally) or an embedding relation. Therefore, coordination bubbles allow the
30
okay so
{
|
what
what changed your mind
Alfred and Bernard fall
}
Figure 4: Disfluency
{
|
and
Alfred
Bernard
Figure 7: Coordination according to Hudson
}
fall
Alfred and Bernard fall
Figure 5: Coordination in a pile
Figure 8: Coordination according to Rosta
[. . . ] pile markers like and or or, usually called coordinating conjunctions,
are in a syntagmatic relation only with
the conjuncts and do not play any role
in the combination of the conjuncts
with the context as they can only appear between two conjuncts (Kahane
and Gerdes, 2009, § 3.1)
1.4
Summary of options
Regarding simple coordination of verbal dependents, differences between models are all linked
to the hierarchical position of the conjunction. The
coordinating conjunction can depend on:
• the coordination relation (Tesnière, 1965);
Formally, bubbles and piles can be combined.
The resulting formalisation displays three sets of
relations: plain syntactic dependencies, in a tree
equivalent to Mel’čuk’s, orthogonal paradigmatic
relations, and pile marking relations (Kahane,
forthcoming). As a result, the analysis of ex. 2
is represented in fig. 6, where solid arrows are regular dependencies, the double line expresses the
paradigmatic link, and the dashed arrows express
the value of the pile marker.
• nothing (Hudson, 2010; Kahane, 1997; Kahane and Gerdes, 2009);
• the first conjunct (Mel’čuk, 1988);
• the first conjunct in a parallel set of dependencies (Kahane, forthcoming);
• the verb (Rosta, 2006).
It can govern:
fall
• nothing (Tesnière, 1965);
Alfred
• [undefined] (Kahane, 1997; Kahane and
Gerdes, 2009);
and Bernard
• both conjuncts (Hudson, 2010; Rosta, 2006);
Figure 6: Coordination with tree sets of relations
• the following conjunct (Mel’čuk, 1988);
• the following conjunct in a parallel set of dependencies (Kahane, forthcoming).
Word grammar. Word grammar has a mainly
semantic definition of dependency: a dependent
makes the meaning of its governor more precise
(Hudson, 2010, 147).
Following most recent formulations of the word
grammar dependency model (Hudson, 2010, 176181), a coordinating conjunction has no governor
and is itself the governor of the conjuncts. These
also depend on the verb. Ex. 2 would thus be analysed as in fig. 7.
Another option (Rosta, 2006, 189-191) would
be to make the conjunction the dependent of the
verb, which would govern each conjunct if there
was no coordination (fig. 8).
As far as the concept of dependency is concerned,
we will retain Mel’čuk’s definition hereafter. This
first choice compels us to reject Tesnière’s description, because a word cannot depend on a relation.
2
Segmental underspecification in OF
OF is the ancestor of Modern French. It can
be roughly described as a V2 analytic language.
Some remnants of Latin nominal declension remain, but they are often too poor to guarantee the
univocity of the form/function relation (Moignet,
1988, 87).
31
may vary for the same dependency relation. For
example, it is well known that the Latin subject
was marked using the nominative case, while in
Modern French, the subject is marked by its position in the clause. Once a specification becomes
tightly bound to the way a function is expressed,
its use becomes compulsory.
Being a major written language from the 11th
century to the 14th century, OF has been well described in several more or less traditional grammars, e.g. Foulet (1968), Moignet (1988), Ménard
(1994), Buridant (2000). However, grammars do
not investigate the syntactic description of coordination phenomena in detail, and their contribution
to the problem is generally limited to a list of coordinating conjunctions and their semantic or discursive values, with the main focus on coordination
of clauses or sentences. More useful is the very
comprehensive study by Antoine (Antoine, 1958;
Antoine, 1962), which examines many aspects of
coordination from a diachronic and a synchronic
point of view, but lacks a proper syntactic modelisation of the structure. However, it contains many
well-classified examples and remains very useful.
We use the concept of specification (section 2.1)
to show that OF has many “segmentally underspecified” constructions (section 2.2). The adequacy of the models can be evaluated with this
property (section 2.3).
2.1
2.2
Segmental underspecification in OF
However, there is never a compulsory segmental
mark for every function. Moreover, marks tend to
be polyfunctional; e.g.:
• nominal structures expressing the semantic
recipient are generally indirect (prepositional
specification with a), but the preposition can
be absent (Moignet, 1988, 296), as in:
(4) Nos avons donet Warnier une mason
“We have given W. a house” (Charter,
1252, 3)
• nominal structures expressing a genitive relation can be specified by the preposition de,
but this specification is not compulsory when
the possessor is a human being, as in la fille le
roi [“The king’s daughter”] (Moignet, 1988,
94);
Minimal relation and specification
concepts
Following Alain Lemaréchal’s work, we assume
that every syntactic relation has an underlying
minimal relation (Fr. relation minimale) that
has hardly any formal mark. Put simply, some
words are connected simply by being used together, without the need for grammatical information other than the part-of-speech class they belong to. For instance, using red and book together will generate an understandable phrase that
“works” (Lemaréchal, 1997, esp. 3 and 103). At
this “minimal” level, the orientation of the dependency relation is not important.
However, languages tend to add grammatical marks that help to distinguish different functions: prosodic marks, segmental morphemes, etc.
The addition of such marks over a minimal relation is called specification (Fr. spécification) by
Lemaréchal (1997, 107-114). Specifications are
generally combined in complex context-dependant
mark sets. The use of marks make the definition of
the relation more precise, and generally allows the
governor of a relation to be identified. For example, it is the lexical verb that controls the form of
its dependents: most constraints over the dependents are stored in the lexicon.
From a diachronic perspective, specification
• subordination is generally marked by conjunction, but parataxis also exists (Moignet,
1988); see also the extensive study by Glikman (2009).
• even when these prepositions and conjunctions are used, they can have multiple meanings (Moignet, 1988).
Hence we claim, following Mazziotta (2009, 149150), that OF can be seen as a language in
which the syntax relies less on segmental specification than on semantic categories and situational/contextual factors. Consequently, models
used to describe OF should not systematically treat
segmental specification morphemes as governors.
2.3
Consequences
The segmental underspecification of many structures in OF has a direct impact on the choice of the
model best suited to describe the language. Given
the fact that grammatical words such as conjunctions and prepositions are, in some cases, optional, grammatical words cannot always be considered as governors of prepositional or conjunctional phrases, because these words do not fully
32
which expresses the semantic AGENT. To kill the
second argument has the form of an object and is
the semantic PATIENT. One can say that to kill can
govern two types of arguments combining a specific form to a specific meaning. Only one occurrence of each argument type can occur in the same
clause. On the other hand, adjuncts are not subject
to such constraints of form, meaning or presence.
For all languages, juxtaposition is the construction that allows speakers to multiply each argument type of one verb or adjuncts. Simultaneously
using arguments of different types (such as a subject expressing the agent and an object expressing
the patient) is not juxtaposition.
determine the passive valencies of these structures (i.e. the way they combine with a governor), which is the prominent criterion in evaluating
directin of dependency (Mel’čuk, 2009, 27-28).
It is quite probable that many grammatica units
are indeed compulsory (Moignet, 1988, 293), but
the dependency description of OF is not complete
enough to state it firmly in every case. It is better
to keep the description at the level of the minimal
relation while dependency remains unclear.
Hence, if we want to investigate such questions
with respect to the coordinating conjunction, it is
important to choose a model in which the hierarchical position of the conjunction remains undefined. At first glance, the bubble-tree and the pile
models, as well as a combination of the two, seem
a perfect fit, because they do not state dependencies regarding the conjunction.
3
Juxtaposed dependents. Orientations 1, 2, etc.
of a verb can thus be duplicated without using any
grammatical device:
(5) Homes, bestes, sont en repos
humans animals are in rest
“Humans, animals are resting” (Antoine,
1958, 561, quoting Eneas, 2163)
Coordination as a specified
juxtaposition or apposition
In this section, we show that there exist two types
of coordination. The first must be considered as
a special case of juxtaposition (section 3.1). Relying on the structural equivalence between juxtaposition and apposition, we will also demonstrate
that the second type of coordination can be seen as
a special case of apposition (3.2).
3.1
(6) Bien li
siet
cele ventaille,
well to him is suited this faceguard
li hiaumes, li escus, la lance
the helmet the shield the spear
“He is well clad with this faceguard, the
helmet, the shield, the spear” (Stein et al.,
2008, BretTournD, 2202)
Specified juxtaposition
The same is true of the adjunct position, which is
naturally unbounded.
Given the possibly underspecified status of coordination, we follow Antoine’s insight, focusing our
survey at first on what one might call “implicit”
coordination, in order not to assign too important
a role to the conjunction initially (Antoine, 1958,
461).
Specification. From our point of view, the coordinating conjunction that can be used between
juxtaposed arguments is a specification device that
is added to a relation that already exists. In other
words, there cannot be a coordination if there is no
multiplication of any type of argument. As a result, although the word et is present in ex. 7, there
is no juxtaposition, and therefore no coordination:
Argument types. Let us first try to define what
one may call juxtaposition at clause level (not between clauses). There may be juxtaposition between dependents of the verb, but what makes juxtaposition different from simultaneous use of different arguments of the same verb?
From a syntactic-semantic perspective, the
verb, as a selected lexical unit, has a predetermined set of valency patterns, constraining the semantic role and the morphosyntactic expression of
its arguments (Lemaréchal, 1989, 102). For instance, in its prototypical transitive use, the verb
to kill has a first argument of which the grammatical form is that of a subject (possible agreement
with the verb, substituability with he, etc.) and
(7) Nos oïemes che ke li veritauele dissent
we heard what the witnesses said
et par serement
and under oath
“We heard what the witnesses declared under oath indeed” (Charter, 1260, 10)
Although et is present, the adjunct et par serement
is not coordinated, because there is no other juxtaposed adjunct in the clause. Therefore, et has
to be considered as a mark of specification of the
33
relation bounding the adjunct to its verbal governor dissent (we will not elaborate on the structural
position of the preposition par here). From a semantic perspective, the word et adds emphasis to
the adjunct.
If the coordinating conjunction is a specification
mark that combines with an already existing relation, the conjunction cannot be the governor of the
second conjunct, nor can it be a third co-head in a
common bubble. If the coordinating conjunction
is secondary, Mel’čuk’s description presented in
1.2 does not hold for OF.
Moreover, following Mel’čuk’s definition of dependency if the conjunction forms a phrase with
the second conjunct and is directly linked in a
dependency relation with the first one, it should
be described as the governor of the second conjunct (Mel’čuk, 2009, 26-27), which cannot be
the case. Therefore, there is no dependency relation between the first conjunct and the conjunction, which must be described as a dependent of
the conjunct following it.
In other words, we also reject the classical assumption that juxtaposition is a coordination from
which the conjunction has been deleted (Tesnière,
1965, ch. 137, § 1). This is a matter of frequency,
rather than of grammatical organisation: specification is more frequent, but it does not mean that it is
more basic from a structural point of view. Fig. 9
shows our simplified analysis of ex. 8.
thogonal relation between (groups of) dependents
is inherited from this model.
Such model integrates seamlessly polysyndeton
(ex. 9):
(9) li
baisse et le bouche
to him DATIVE kisses and the mouth OBJ
et le nes
and the nose OBJ
“He kisses him on the mouth and on the
nose” (Stein et al., 2008, ElieB, 2599)
Here, the first coordinating conjunction depends
on the first conjunct, as shown in fig. 10.
DAT- OBJ
li
la bouche
le nes
et
et
Figure 10: Representation of polysyndeton
Indeed, there are many simple examples of the
specified construction in OF. According to our
knowledge of this language, and to the texts we
have read so far, we have found that juxtaposition
is very often specified in the case of a coordination of genuine arguments (which excludes coordination of adjuncts). We believe that in the written language underspecification tends to become
rarer over time (a diachronic survey would be necessary). Note that adjuncts are obviously not subject to this emerging constraint.
3.2
prenez
mon escu
OBJ
OBJ
(8) Prenez mon escu et ma lance
“Take my shield and my spear” (Defourques and Muret, 1947, Béroul,
v. 3586)
OBJ
baisse
Specified apposition
We claim that coordination can also be a specified case of apposition, which is a property of OF
but not modern French – Bilger (1999, 263-264),
among others, gives no example of specified apposition.
OBJ
ma lance
et
Comparing apposition and juxtaposition. Intuitively, appositions are generally described as simultaneous expressions of the same object; e.g.:
Figure 9: Juxtaposition specification
The coordination relation takes the form of a
bubble and the conjunction depends on the second conjunct. The juxtaposition has duplicated
the object-PATIENT argument of the verb without
changing its valency. Note that the model is not
exactly a bubble tree, because dependency relations cannot normally cross the border of a bubble, but the main idea of coordination being an or-
(10)
34
Li enemy,
li aduersaire dunc
the enemy warriors the opponents then
se desrengent
are restless
“Then, the foes are restless” (Stein et al.,
2008, EdmK, 2065)
to the same object, they are said to be coreferent.
For instance, a noun and the pronoun replacing it
are coreferent. Coreference is a major semantic
characteristic of apposition, distinguishing it from
juxtaposition: apposed nouns share the same designatum. Note that subject/verb agreement cannot
be considered as a reliable grammatical mark of
the difference between apposition and juxtaposition (Foulet, 1968, 201-202).
(11) Tu vouloies ceste angoisse, ceste dolor,
You wanted this anguish this pain
ceste painne pour nostre amor [. . . ]
this mourning for the love of us
“You wanted to experience this pain for
our sake” (Stein et al., 2008, PassJonglGP,
497)
(12) Adont m’
arés
vous retenu
Then me OBJ will have you retained
a vostre ami, a vostre dru
as your lover as your lover
“Then, you will hold on to me as your
lover” (Stein et al., 2008, JacAmArtK,
1972)
Specification. The apposition relation can be
specified by the use of a coordinating conjunction,
as seen in ex. 1, and in the following excerpt.
(13)
Tesnière has the following insight:
The form of the junction line is identical
to the form of the apposition line, since
both are horizontal (Tesnière, 1965,
ch. 136, § 5, our translation)
Poor en ont tuit et esfroi
Fear of it have all and fright
“They are all afraid of it” (Defourques and
Muret, 1947, Béroul, 1722)
Since we consider juxtaposition and apposition to
be syntactically equivalent, our analysis of paier
cel pris et celle summe is shown in fig. 11, where
the dashed line represents the coreference relation.
But he argues (Tesnière, 1965, ch. 69, §§ 5-6 and
ch. 139, § 6) that the apposed node, even if it is
bound by an horizontal line, remains dependent
upon the node to which it is apposed (the relation that unites them is a connexion). Underlying
his argumentation is the assumption that apposition is not a clause-level relation: apposed nouns
are governed by a node that may be an argumental dependent. This may be true, but there is a
major difficulty in determining what is apposed to
what. Moreover, apposed dependents of the verb
share the same constraints bound to their function
(e.g. the use of the preposition a in ex. 12).
It is often not possible to decide which apposed
word would be the governor in an apposition relation. As they share the same argument type, apposed words have the same passive valency, and
therefore would trigger the same agreement in the
same context. From a semantic point of view, they
are lexical synonyms (enemy/adversaire in ex. 10
or ami/dru in ex. 12) or they refer to the same object or fact (angoisse/dolor/paine in ex. 11). The
hierarchy remains undefined.
The difference between argumental apposition
and juxtaposition is only semantic – the fact has
been highlighted by Blanche-Benveniste and Caddéo (2000) for spoken modern French, and by
Touratier (2005, 290) in a constituent-based approach – as it is a case of coreference (Hudson,
2010, 229-232). Where several dependents refer
OBJ
cel pris
paier
OBJ
celle summe
COREF
et
Figure 11: Specified apposition
Contrary to juxtaposition, we suggest (again,
this should be verified), underspecification has
generalised in apposition over time. Note that
modern French can still specify appositions when
they are not directly dependent on the verb. Thus,
14 is grammatical (the unique determiner implies
that there is only one noun phrase), but 15 is not:
(14)
Je vois ma chère et tendre
“I see my dear and sweet”
(15)
**Je vois ma chère et ma tendre
3.3
Conclusion
As far as verbal dependents of OF are concerned,
coordination is a form of juxtaposition or apposition that is specified by the use of a coordinating conjunction. The fact that apposition can be
specified in the same manner as juxtaposition is a
property of OF that has not survived into modern
French.
35
Since both constructions occur without this
specification, the coordinating conjunction has to
be described as a dependent of the conjunct following it. Of course, this position of the conjunction should be reserved to languages where its
presence is not compulsory: where the conjunction is mandatory, it has the position of a governor.
However, according to Caterina Mauri (2008, 60),
juxtaposition without specification is always possible at clause level, in all languages she has investigated:
Claude Buridant. 1977. Problèmes méthodologiques
dans l’étude des traductions du latin au français
au XIIe siècle: le domaine lexical. Les couples de
synonymes dans l’histoire de France en français
de Charlemagne à Philippe-Auguste. In Danielle
Buschinger, editor, Linguistique et philologie: Application aux textes médiévaux. Actes du colloque
des 29 et 30 avril 1977, Champion, Paris. 293–324.
Claude Buridant. 1980. Les binômes synonymiques.
Esquisse d’une histoire des couples de synonymes
du moyen âge au XVIIe siècle. Bulletin du centre
d’analyse du discours, 4:5–79.
Claude Buridant. 2000. Grammaire nouvelle de
l’ancien français. Sedes, Paris.
Asyndetic constructions consist of the
simple juxtaposition of the two SoAs
[i.e.: ‘states of affairs’, “hyperonym
for the words ‘situation’, ‘event’, ‘process’ and ‘action’” (Mauri, 2008, 32)],
and the specific coordination relation
existing between them is inferred from
the context of communication and from
their semantic properties. Asyndesis
is always possible and occurs in every
language as a more or less stylistically
marked strategy.
Charter. 1252. 1st of March 1252. Archives de l’État
à Liège (Belgium), couvent de Robermont.
Charter. 1260. 9th of May 1260. Archives de l’État à
Liège (Belgium), couvent du Val-Benoît.
Charter. 1278. 1st of August 1278. Archives de l’État
à Liège (Belgium), collégiale Saint-Denis.
L. M. Defourques and E. Muret, editors. 1947. Béroul.
Le roman de Tristan. Poème du XIIe siècle. Number 12 in CFMA, Paris, Champion, 4th edition.
Lucien Foulet. 1968. Petite syntaxe de l’ancien
français, Paris, Champion, 3rd edition.
It means that the dependent position of the conjunction can be generalised in the case of juxtaposition.
Paul Garde. 1981. Des parties du discours, notamment
en russe. Bulletin de la Société de Linguistique de
Paris, 76(1):155–189.
Julie Glikman. 2009. Parataxe et Subordination
en Ancien Français. Système syntaxique, variantes
et variation. Université Paris Ouest Nanterre La
Défense and Universität Potsdam. Thèse de doctorat.
Acknowledgements
We would like to thank Thomas M. Rainsford,
Julie Glikman, Brigitte Antoine, Sylvain Kahane
and Lene Schøsler for proofreading and content
suggestions.
Richard Hudson. 2010. An introduction to word grammar. Cambridge Textbooks in Linguistics. Cambridge University Press, Cambridge.
Sylvain Kahane and Kim Gerdes. 2009. Speaking in
piles. Paradigmatic annotation of a French spoken
corpus. In Proceedings of Corpus Linguistics Conference 2009, Liverpool, 16 pages.
References
Gérald Antoine. 1958. La coordination en français.
Tome I. D’Artrey, Paris.
Sylvain Kahane. 1997. Bubble trees and syntactic representations. In T. Becker and H.-U. Krieger, editors, Proceedings of the 5th meeting of Mathematics
of Language (MOL 5), Saarbrücken, 70–76.
Gérald Antoine. 1962. La coordination en français.
Tome II. D’Artrey, Paris.
Mireille Bilger. 1999. Coordination: analyses syntaxiques et annotations. Recherches sur le français
parlé, 15:255–272.
Claire Blanche-Benveniste and Sandrine Caddéo.
2000. Préliminaires à une étude de l’apposition dans
la langue parlée. Langue française, 125(1):60–70.
Sylvain Kahane. forthcoming. De l’analyse en grille
à la modélisation des entassements. In Sandrine
Caddéo, Roubaud Marie-Noëlle, Magali Rouquier,
and Frédéric Sabio, editors, Hommage à Claire
Blanche-Benveniste. Aix-en-Provence, Publications
de l’Université de Provence.
Claire Blanche-Benveniste and Colette Jeanjean. 1987.
Le français parlé. Transcription et édition. Didier
érudition, Paris.
Alain Lemaréchal. 1989. Les parties du discours. Sémantique et syntaxe. Linguistique nouvelle. Presses
Universitaires de France, Paris.
36
Alain Lemaréchal. 1997. Zéro(s). Linguistique nouvelle. Presses universitaires de France, Paris.
Caterina Mauri. 2008. Coordination relations in the
languages of Europe and beyond. Number 42 in
Empirical approaches to language typology. Mouton
de Gruyter, Berlin and New York.
Nicolas Mazziotta. 2009. Ponctuation et syntaxe dans
la langue française médiévale. Étude d’un corpus
de chartes écrites en français à Liège entre 1236 et
1291. Number 354 in Beihefte zur Zeitschrift für
romanische Philologie. Niemeyer, Tübingen.
Igor Mel’čuk. 1988. Dependency syntax: theory and
practice. State University of New York, Albany.
Igor Mel’čuk. 2009. Dependency in natural language.
In Alain Polguère and Igor Mel’čuk, editors, Dependency in linguistic description, John Benjamins,
Amsterdam and Philadelphia, 1–110.
Philippe Ménard. 1994. Syntaxe de l’ancien français.
Études médiévales, Bordeaux, Bière, 4th edition.
Gérard Moignet. 1988. Grammaire de l’ancien
français. Morphologie – Syntaxe. Number 2 in Initiation à la linguistique. Série B. Problèmes et méthodes, Paris, Klincksieck, 2nd edition.
Andrew Rosta. 2006. Structural and distributional
heads. In Kensei Sugayama and Richard Hudson,
editors, Word grammar. New perspectives on a theory of language structure. Continuum, London and
New York, 171–203.
Achim Stein, Pierre Kunstmann, and Martin-Dietrich
Gleßgen, editors.
2008.
Nouveau Corpus
d’Amsterdam. Corpus informatique de textes littéraires d’ancien français (ca 1150-1350), établi
par Anthonij Dees (Amsterdam 1987), Stuttgart, Institut für Linguistik/Romanistik, 2nd edition. DEAF
identifiers.
Lucien Tesnière. 1965. Éléments de syntaxe structurale. Klincksieck, Paris, 2nd edition.
Christian Touratier. 2005. Analyse et théorie syntaxiques. Publications de l’Université de Provence,
Aix-en-Provence.
37